Monthly President’s Message

What will the future of forecasting look like?
Gail Hartfield, July 2017

When I was a kid, my family, like many at that time, had a hefty set of encyclopedias that I relied on for school research projects. It was laborious at times, thumbing through one or more of the 24 volumes to find the information I needed. Back then, we could not have foreseen a day in which tools like Google and Wikipedia could provide instant information at our fingertips. And the idea of phones as pocket computers during the era of rotary dialing? Also an inconceivable notion for most of us at the time. Such innovations often start as ideas long before they become reality, and advances in the science of weather forecasting have been no different. Many meteorological innovations— including radar, modeling, and satellites— began as concepts or hopes a few decades prior to their creation.

There’s no doubt we’ll see new techniques in forecasting, observing, and warning in the coming years and decades. But what exactly will these innovations look like? Can we get a clear picture of what our future will bring in 10, 20, or 30 years? Or are such advances difficult to envision right now? I decided to polish up my own crystal ball to catch a glimpse of what we might see in the future:

● Better model initializations, better data assimilation, better physics, better computing, and higher temporal and spatial resolutions. This one is a certainty since we’re already seeing vast improvements in modeling every year with ever-growing computing power. But how accurate will we get? Will we depict exactly when and where the sea breeze will move? Will we regularly and reliably model individual updrafts? How about 3-D modeling with detailed and realistic depictions of the vertical extent of convective cells? Or precise projections of wintertime hydrometeor phases?

● Better warnings and very short-term forecasts. The advances in modeling just mentioned would lead to this. We’re already seeing an example with the experimental Warn-On-Forecast (, which combines radar data and other observations with very high-resolution model output to create successful warnings with an hour (or more) of lead time. Efforts to connect observations with highly detailed and accurate simulations are likely to increase in the coming years given the growing focus of the weather enterprise on decision support services. And operational meteorologists need to be ready to accommodate this altered paradigm within warning operations.

● Greater reliance on models, especially at longer ranges. Around 25 to 30 years ago, forecasts for Day 3 and beyond were coarse and generalized, with little detail and very few model-as-source options. Now, numerous models are at our disposal, generating projections of sensible weather like hourly precipitation and 2-meter temperatures many days out (although accuracy can vary, of course). As these models improve, the role of human forecasters in the extended range will evolve into one in which we become focused on opportunities for model corrections, particularly for significant weather events, while spending less time constructing the forecast itself. To be sure, the potential for a changed or reduced human role could be difficult for some to deal with. But these changes should allow forecasters more time to focus on the aforementioned short-term forecasts, providing more temporal and spatial detail and serving a wider range of user needs.

● Working with the “weather app” culture instead of against it. When the average Joe and Jane Public want a forecast, they want it now. Nothing complicated, just the basics of what the weather will be like. A cute picture and a number. No details…except when they want to know exactly what will be happening at the 4 p.m. soccer game tomorrow. (Oh, and we need to be correct, too!) How can we possibly provide people with the detail and simplicity that they want? The weather enterprise has already started to figure this out, with forecasters in every sector working to provide both kinds of forecasts in one display without sacrificing scientific integrity. But we have a long way to go to provide an accurate, scientifically sound forecast in an easily digestible way that allows people to quickly get their forecast, make their decisions, and go. This will require forecasters to increase their understanding of how people gather and process weather information. Not an easy task, but if we can’t beat ‘em, we’ll have to join ‘em.

● Satellite data explosion. The incoming generation of satellite technology, spearheaded by GOES-16, is merely the beginning of a remote sensing mini-revolution. Meteorologists are just now digging into all that is (and will be) possible with the new GOES-16 instrumentation and products. And many more tools on the way, including aircraft icing threat and ocean current detection, are bound to improve and expand our forecasting capabilities. As researchers and forecasters explore these new datasets as applied to real-world problems, forecasters will find new ways to use this bounty of information.
Of course, these are not all of the changes we may see in the future. What do YOU think forecasting and warning services will look like in a few decades? Feel free to message us with your thoughts on the NWA Facebook page.

Comments are closed.